 Hey folks In my amazing Competence I once again left the microphone muted and yapped for hmm looks like about 10 It's so we're gonna start it again, and we'll just see how it comes out We've got a tremendous amount of stuff to talk about so here. Just we go The scientist asked if he could replace BV's brain and BV said okay, and so the great scientific experiments began It wasn't always a smooth road Sometimes BV just lost interest in the ball The doctors suspected BV was suffering from post-brain replacement depression So the doctors ran lots of tests to see what was going on It turned out that the problem was a communications failure in the spine the connection between BV's body and BV's new external brain But the engineers got to work and soon BV was feeling better Sometimes it seemed like BV was even more exuberant about chasing the yellow balls Which the scientists had decorated for mystery on ex-scientific purposes And the engineers snaked the spinal cord farther and farther from BV's body and kept attaching it to new brains Eventually they put a single oolahm atom in control and not long after that It was a network of cooperating atoms routing signals in and out of BV's brain It was beautiful, but still still in all BV couldn't change its own mind and that would have to change Okay, that's That's the story of BV The goals for today to have BV's brain coated new alarm running on the t2 matrix It is coated in oolahm at least a bit. It's a beginning. It's a legitimate beginning But it's not running on the t2 matrix. It's running on the simulator The main one was that BV was supposed to make changes to its own wiring that has not happened And that is going to be the main goal for the next update and we're getting down to the end of the brain challenge so BV's brain what one just one point I wanted to try to make quickly was that Once their actual spatially distributed brain with sensors Signals that arrive in one place and motor stuff that has to that somewhere else and the signals have to get routed From one to the other it does create a significant lag in terms of the number of sort of Fundamental steps of the simulate the world simulation between the time that a sensor is detected and the corresponding motor command is issued And you can see it. Well, at least I can see it In you know, BV's wagon is but around BV's wagging its butt around More because of the lag in tracking, but it's still working out. Okay And in fact, I want to make it even slower by putting more complex computations and stuff going on in there So this is where we've got it. We've got it all the way up to the top now Except it's not it's not running on MFM t2 running on the t2 tiles. It's running on MFMS the simulator But next month for sure Okay, and the schedule is you know, here we are next month is supposed to be the sensory motor Homunculus now not even exactly clear. What the heck that is or what the heck that would mean in a Spatially distributed brain design We'll see I was I was hoping to have a little bit more experience with the brain stuff going by now But okay, so be it the one thing I do know what I want to see on April 2nd, which is actually fairly short delayed in the next update is It's based on a subsumption Architecture the subsumption architecture was a term introduced by Roddy Brooks back in the 90s When he was building robots and it was the idea that you build robot behaviors from the bottom up first You build basic things like be able to balance yourself or whatever and then you build a layer on top of that that allows you to Avoid obstacles and you build a layer on that to build maps or whatever it is and you work your way up and the point is is that the lower levels are just working automatically so that if the Mapmaking doesn't have anything it wants to do it doesn't do anything and the lower level things continue to do their operations So the difference between traditional computing is that the lower level things in a computer do nothing Until the CPU comes in and says okay increment your number by one or whatever it is Subsumption is very different because you have these active processes running below you that you modify you steer that kind of thing And whatever we get for for BV. I want to be able to explore the subsumption architecture in Ulan. We'll see how it goes So oh education and outreach. Yeah So once again Last year of the International Society for Artificial Life I saw gave me this award for Education and outreach for the t2 tile project for these exact videos That we're doing here right now And and that was nice And I sort of thought there was supposed to be this physical award that came with it But it didn't seem that there was and so I didn't know exactly what was going on But just in the last week the award finally arrived it and it's it's great It's you know, these are that's all built out of metal it's these are all nails that have been pounded and bent and Into this very organic Lifelike-looking thing made out of metal made out of nails. This is done by a guy named Cliff Bohm who's a hacker and a thinker and artist in in Michigan and I This is great. I'm really happy to own this Thanks to the ice ala wards committee for thinking of me and thanks to Cliff Bohm You know, so so here it is and You know, this is the sort of thing that you I don't think you'd really want to have it on the wall when you're having Like a big argument with your business partner because you know, someone's gonna get hurt Okay, so that's that Global phase war man, I threw more than a half of the thing dear Okay This I'm calling it a side quest. This is not a word that I use a lot I don't play games that have quests really so But this is something that goes back a lot of time. So I want to kind of catch up and start doing it So from over a decade ago This was the demon horde sort video that was a lot of people's first introduction to Robust first computing and you have these blue data items that are come in at the right-hand side And they get pulled through the matrix by the demon horde the red Sorting atoms and then they get extracted at the left-hand end by the black output grid and you know, it was it was very interesting It worked Robustly It has this weird property. It's so difficult for us to get past, you know, absolute correctness and all but you know This is the whole story I'm preaching to the choir. Hopefully for the folks who are here One of the things that always bugged me about this was that like the input grid just went vertically up and down until It hit the end of the universe and the same thing with the output grid and the demon horde sort just grew and grew and made more of itself For it took as much space as there possibly was and so, you know That's fine, but you're never going to be able to use a demon horde sorter as a component of Something else unless you can figure out a way for it to manage its own size So I was always that was always bugging me But then, you know over the years did lots more work. So like by last year we did the digital replicator and Here is the hc3 grid a 3x3 grid of atoms That knows exactly how big it wants to be it can grow it can manage to avoid running into other grits and so forth exactly so that They can share components that you can have multiple components sharing the same grid So the question was, you know, okay, if we've got these input signals for the Breitenberg vehicle, you know Left yellow and right yellow and we've got these motor Terminals left motor and right terminal and motor and we want to say, you know cross them to make the to be Breitenberg vehicle model How are the signals for the light sensors going to know where the motor terminals are and so the idea was well We could build a grid we could build a grid out of like hc3 for example And make radiance so that I you could say I'm one hop away from the left motor I'm ten hops away from the right motor and all of the grid would figure out these numbers And then the signals could just follow the gradients to get where they needed to go And that is what we have done and it works fine. Let's take a look at some alright, so So yeah, what did you see that that flash there, you know What the heck was that that flash that started that it was the same thing in slow motion that it Whatever That was the Brain of booting itself up by building a grid Out so that the gradients could form so the signals could find their ways to the motor terminals How did the grid know how big to make itself? Well, we're back to the demon hordes or it didn't it just it just fills space Because you know say well, we only got one brain. So go ahead fill space But then there's this problem, right? a so the hc3 grid whoops here we go the hc3 grid is 89% nothing that was a title of one of previous updates and Because it for every three by three nine total sites. It's only got one occupied and eight empty That but that means there are nine different positions that a three by three grid could be based at based on where that one Anchor thing is based on so what happens if for whatever reason we have part of the grid growing with phase One and part of the grid growing with phase eight or whatever it is and they run into each other what happens So that's what we've got here. So Let's take a look at this Well, we didn't look at C2 first Right So here is the grid just growing itself and you know, it works fine And these things are changing colors. They're going from red to yellow to white because gradually they're gonna know they're sort of storing up energy that kind of thing and So here That's just so let's see to him and now we want C3 Okay, I'm all discombobulated So here I've seated a bunch of these little grids and it ended up having two different phases. So look at this You can see right here The two different phases are running into each other and and they're having a war Basically because when they see one that's on the wrong phase, they erase it and when they see empty spots They put themselves in and it just goes back and forth and in fact It actually follows the the battle lines between two different phases tends to fall on Boundaries between the T2 tiles, it's like rivers and geographic boundaries where It's easier to defend when you have the the other side has to cross a Difficult boundary before they can attack you but we're getting nowhere here So I did a whole bunch of work on it to figure out ways So what we really would like to do is have whichever we don't care which phase work wins It's all gonna come out the same in the end We just wanted to have one so that we can do our gradients and get things going So, you know, if we don't care who's gonna win The obvious thing is to pick whichever side is bigger because once a Single phase takes over the entire universe. It's gonna be the biggest thing there is So we might as well start with the thing that's already bigger and let the rich get richer and Eliminate the other ones which is difficult because I always feel like I want to be sitting for the underdog, but gotta go here And this doesn't work But what we can do is we can do like a battle where the hinterlands are producing resources weapons energy something Which they then give they ship it forward to the battle line And so whichever side has better Manufacturing in logistics to deliver resources to the battle line is gonna tend to win more battles And that's what we've actually got here And So let's see it Okay, so that was four different phases that well, it still is four different phases But the two on the diagonals are getting squeezed out now. We've got just the two on opposite sides and Right there. Yeah, so the battle line is gradually getting pushed toward the Northwest And you know, once the manufacturing base starts getting attacked We can it goes very quickly like that And so there's an example where it works. Well, we are running long. So I am going to jump ahead to and this is now running on the teacher matrix and The sad fact of it is is We can get to the point where All you know, if you've got a whole bunch of different phases, they'll sort themselves out. Look at this And Again four phases that gradually squeeze out to two and then in this case it actually succeeds in finding one But in the next case Almost the same if they're close enough to each other and the battlefield is big enough It will not converge and so here's a case once again the rich get richer All of this stuff where there's no white is, you know, that's all maxed out. It's it's ready to go and And there's one more and this is the one that looks very similar But for whatever reason the battle line between them it is Just enough balance, they're just perfect against each other and this battle line never ends It's been running for days and this This is the sad outcome right that Any given logistics and resource system is going to have a certain amount of maximum Flow that it can bring to the front and by whatever that maximum flow is then any more Stuff that it's got further out doesn't have an impact on what happens on the border This is just gonna go on indefinitely and that's what we don't want We want to have stuff that'll manage its own controls and then Leave room for other stuff. So that's it Okay, so I've gone long The goal for next time BV's distributed oolong brain running on the t2 matrix And and it changes its mind using the subsumption architecture subsumption style of computing somehow and once again have fun and Remember to unmute the microphone. Let's not go through in a row That's it. That's so much. Thank you so much for stopping in and I hope to see you next time